Given a two-way crossover network where the low-pass filter for the woofer is 1st order, and the high-pass filter for the tweeter is second order, what should the crossover frequency be? Would it be the frequency at which the total attenuation of roll-off of the woofer itself plus that effected by the low-pass filter equals -6 dB? Or should the two cut-off frequencies be unequal according to some design method?
I have quite a few books that discuss crossover networks, and I've checked the ESP site, but there is no treatment of mixed order crossovers there. Possibly the only way to optimize a mixed order network would be by testing it?
As I'm analyzing a commercial loudspeaker system and I don't have the full specifications of the drivers of it, I'm unable to simulate the design to see if it has the correct cross-over frequency.
Regards,
Pete
I have quite a few books that discuss crossover networks, and I've checked the ESP site, but there is no treatment of mixed order crossovers there. Possibly the only way to optimize a mixed order network would be by testing it?
As I'm analyzing a commercial loudspeaker system and I don't have the full specifications of the drivers of it, I'm unable to simulate the design to see if it has the correct cross-over frequency.
Regards,
Pete
Are you talking the electrical or the acoustical order of the crossover?
If you are talking electrical order, that can be almost any order on the acoustical side (depends on the acoustical and electrical responses of the drivers, combined with the electrical filter).
If you are talking acoustical order, then which 2nd order response? Butterworth, Bessel, Linkwitz-Riley etc?
I think the best you can do is to modell your idea in a crossover modelling software.
Anyway, what's matter from acoustical point of view is the acoustical order of the crossover.
If you are talking electrical order, that can be almost any order on the acoustical side (depends on the acoustical and electrical responses of the drivers, combined with the electrical filter).
If you are talking acoustical order, then which 2nd order response? Butterworth, Bessel, Linkwitz-Riley etc?
I think the best you can do is to modell your idea in a crossover modelling software.
Anyway, what's matter from acoustical point of view is the acoustical order of the crossover.
Ohh, and the crossover of a speaker cannot be simulated from driver specs and electrical order of the filter, it needs to be measured acoustically and electrically.
I'm referring to electrical order. At the moment I can't say as to the type of the second order high-pass filter. But if for example the acoustic response of the woofer remains flat at frequencies much greater than the crossover frequency, should both high-pass and low-pass filters produce - 6 dB attenuation at the crossover frequency, or -3 dB? This question is based on the fact that usually the electrical roll-off at the crossover frequency for odd order and even order filters equals respectively -3 dB and -6 dB.
Thanks,
Pete
"What frequency?" It depends on the drivers!
Anyway, you need to measure the drivers and load the data into a network simulation program and figure out the crossover. A trick to get pretty good phase match is second electrical on the woofer and third on the tweeter. Sometimes these can work out to LR4 acoustic pretty close. But it depends. So model prototype and measure. Repeat.
Ball park, very few 1 inch domes should be used below 2K. Very few 6 incher mid-woofers can be well behaved any higher, so for a generic rule of thumb, reasonably well behaved drivers in a 2-way bookshelf cross in the 1800 to 2200 range. Some tweeters like the XT-25 really need to be 2500 or higher. Some SB can squeak in on the low side. Aluminum cone mids need lower, steep crossovers due to breakup.
FWIW, first electrical on a woofer is way too shallow. Not only do woofers not behave that well, but phase differences into the midrange will be terrible. Bad imaging. Second on the tweeter will often push it too close to excursion limits and excessive distortion. Often to get away with low order filters you wind up adding notches and shelves where the parts count is higher than if you used a proper filter to start with.
Anyway, you need to measure the drivers and load the data into a network simulation program and figure out the crossover. A trick to get pretty good phase match is second electrical on the woofer and third on the tweeter. Sometimes these can work out to LR4 acoustic pretty close. But it depends. So model prototype and measure. Repeat.
Ball park, very few 1 inch domes should be used below 2K. Very few 6 incher mid-woofers can be well behaved any higher, so for a generic rule of thumb, reasonably well behaved drivers in a 2-way bookshelf cross in the 1800 to 2200 range. Some tweeters like the XT-25 really need to be 2500 or higher. Some SB can squeak in on the low side. Aluminum cone mids need lower, steep crossovers due to breakup.
FWIW, first electrical on a woofer is way too shallow. Not only do woofers not behave that well, but phase differences into the midrange will be terrible. Bad imaging. Second on the tweeter will often push it too close to excursion limits and excessive distortion. Often to get away with low order filters you wind up adding notches and shelves where the parts count is higher than if you used a proper filter to start with.
The electrical order/filter response is one thing, the other thing is the raw acoustical and electrical response of the mounted drivers, these are combined together and the result is the acoustical rolloffs whose determines the acoustical order of the crossover and that is what you hear, not the electrical order of the filters. You need to measure, otherwise you don't know the acoustical order of the crossing.
Pete, did you finish your studies at university? I remember helping you analyze a simple circuit for which you had to prove that phase angle between two vectors was 90 deg.
You must be confusing me with someone else, as unfortunately I graduated from college decades ago.
I did not confuse you with anyone, and you did start a short thread 8 years and 2 months ago where at least 2 experts contributed. Your statement that you were not a student back then is noted.
The goal is usually to control what you get at the speakers. It usually doesn't matter what filters you use, and typically they will be differrent in each situation.
You are absolutely correct that that was me 8 years and 2 months ago. Even then 😵 I was well past getting my liberal arts degree, and studying electronics on my own. Phasor algebra is still interesting to me (doing the actual math), but right now I'm fairly rusty at it. There are some things that I'm good at, such as borrowing some of the insights of Earl Geddes to design the best home stereo speakers available on the planet. What I'm not good at is retaining what I have learned to do in the past.
Thanks for reminding me of that thread. And yes, several experts contributed to that thread, I'm shocked!
I always think a concrete example simplifies ones thinking. First order plus second order.
Here the splendid Monitor Audio R300-MD speaker designed by Robin Marshall:
A well-behaved 8" ELAC bass combined with a 3/4" SEAS 19 TAF/G metal dome tweeter.
What Robin did was the following crossover:
Ended up looking like this physically:
Sold by the bucketload. Of course, I can do better. But not at the price.
Here the splendid Monitor Audio R300-MD speaker designed by Robin Marshall:
A well-behaved 8" ELAC bass combined with a 3/4" SEAS 19 TAF/G metal dome tweeter.
What Robin did was the following crossover:
Ended up looking like this physically:
Sold by the bucketload. Of course, I can do better. But not at the price.
Here's a two way I did a while back. Sounds great - very clean and dynamic with classic 'BBC' vocals.
CO: Tweeter - 2nd order electrical; bass - 1st order electrical with a notch filter for the L18's 7Khz resonance.
Frequency response:
Reverse nul:
CO: Tweeter - 2nd order electrical; bass - 1st order electrical with a notch filter for the L18's 7Khz resonance.
Frequency response:
Reverse nul:
@system7
The coils of the crossover network of that R300-MD speaker are not oriented at a right angle and it appears that they are maybe 5 cm apart. Given that, isn't there some interaction between their magnetic fields, maybe enough to generate distortion?
@Zuhl
That 2 way has very flat frequency response, much better than that of my "best on the planet" loudspeaker system, also a 2-way. That is also quite an unusual design with a first order high-pass and 2nd order low-pass filter.
The coils of the crossover network of that R300-MD speaker are not oriented at a right angle and it appears that they are maybe 5 cm apart. Given that, isn't there some interaction between their magnetic fields, maybe enough to generate distortion?
@Zuhl
That 2 way has very flat frequency response, much better than that of my "best on the planet" loudspeaker system, also a 2-way. That is also quite an unusual design with a first order high-pass and 2nd order low-pass filter.
Drivers are measured. Put them into XSim and play around with the values until you get - the right xover frequency, a flat frequency response and a deep reverse nul.
Plus 40 years of practice... ;p
I can get the response flatter with a different tweeter - SEAS 22 TAF/G
Needs a zobel though to tame the top end
And here it is...
Plus 40 years of practice... ;p
I can get the response flatter with a different tweeter - SEAS 22 TAF/G
Needs a zobel though to tame the top end
And here it is...
Last edited:
If you have a 1st-order low-pass filter, and a 2nd-order Linkwitz-Riley high-pass filter in a crossover network, what you can do is make attenuation of the 1st-order filter equal to -6 dB at the crossover frequency. Then also the component values of the 2nd-order filter remain what they normally would be. As in an unmixed 2nd-order 2-way crossover network, the tweeter and woofer are oppositely connected to the respective filters.
Phase of the output of the LP filter (voltage drop across R1) equals -60 deg. whereas it would optimally be -90 deg., but in the simulation that I'm attaching the summing remains very flat anyway. In the simulation, I made the crossover frequency equal to 2 kHz.
This is of course considering the electrical side only, but should be good to obtain the value of the reactive component in the 1st-order filter close to what it would best be, I would think.
Regards,
Pete
Phase of the output of the LP filter (voltage drop across R1) equals -60 deg. whereas it would optimally be -90 deg., but in the simulation that I'm attaching the summing remains very flat anyway. In the simulation, I made the crossover frequency equal to 2 kHz.
This is of course considering the electrical side only, but should be good to obtain the value of the reactive component in the 1st-order filter close to what it would best be, I would think.
Regards,
Pete
The woofer coil is often further from the listener. When this is done without measurement you may consider this in view of the more shallow rolloff. For example, 3cm of distance would make up 60 degrees at 2kHz.